Extract Structured Data with Jev and Deterministic Code

What you’ll build

Find candidate values with parsers, use Choice to select the right field, and normalize without inventing identifiers. The result is a runnable decision pipeline, with a fixture mode for checking local behavior and a live mode for evaluating your own TypeSafe account. It prints a recommendation without performing external side effects.

Architecture

Parser → candidate IDs → Jev semantic selection → exact original value → deterministic normalization

Email, phone, money, and IDs are exact values. Jev chooses among supplied candidates; it does not generate a new identifier. Preserve the original string and normalize only what the field’s contract permits. For email, the example lowercases the domain and preserves the local part.

Prerequisites and sample input

Use Python 3.10+; these standalone examples use only the standard library. Live evaluation also requires TYPESAFE_API_KEY in the environment. Download the complete script below rather than copying disconnected fragments.

{
  "message": "Invoice contact: [email protected]. My personal email is [email protected].",
  "candidates": {
    "candidate_0": "[email protected]",
    "candidate_1": "[email protected]"
  }
}

Step 1 — Prepare the evidence

Keep the input shape stable and distinguish verified application facts from user claims. Email, phone, money, and IDs are exact values. Jev chooses among supplied candidates; it does not generate a new identifier. Preserve the original string and normalize only what the field’s contract permits. For email, the example lowercases the domain and preserves the local part.

Step 2 — Define the atomic questions

{
  "billing_email": {
    "type": "choice",
    "instructions": "Which candidate is the invoice contact email? Return none if it is absent.",
    "criteria": {
      "candidate_0": "[email protected]",
      "candidate_1": "[email protected]",
      "none": "No invoice contact among candidates"
    }
  }
}

The question names map outputs back to your code. They are not inference instructions. Put the actual judgment in instructions, and use criteria for category descriptions or ordered levels.

Step 3 — Read answers and apply a gate

The standalone script checks required answer keys, types, allowed categories, numeric ranges, and confidence. Missing or malformed data stops the decision path. The following action policy uses educational thresholds; none have been measured on your data.

def decide(answers, state):
    answer = answers["billing_email"]
    if answer["confidence"] < .9 or answer["choice"] == "none":
        return {"route": "review", "value": None}
    selected = state["candidates"].get(answer["choice"])
    if selected is None:
        return {"route": "invalid_candidate", "value": None}
    local, domain = selected.rsplit("@", 1)
    return {"route": "selected", "value": local + "@" + domain.lower(),
            "deliverability_verified": False}

Step 4 — Keep execution separate

The program prints a route or candidate result. A real executor must apply its own permissions, validation, idempotency, and confirmation requirements. A model label is evidence for a decision, not authorization to perform a consequential action.

Run it

Download the complete structured-data-extraction.py program. Then run:

python structured-data-extraction.py
# After configuring TYPESAFE_API_KEY, opt in to a live billed call:
python structured-data-extraction.py --live

Fixture mode makes no network request and requires no credential. Live mode makes a single call with a 30-second timeout. It does not silently retry or execute any downstream action. For a production queue, add a bounded retry policy for transient failures and a durable review destination.

Expected fixture result

{
  "route": "selected",
  "value": "[email protected]",
  "deliverability_verified": false
}

This output is deterministic fixture data, not a measured Jev response. A live model may produce different values and routes. Keep the full returned probability distributions when diagnosing that difference.

Read the probabilities

A Choice winner alone does not reveal ambiguity. Compare its confidence and distribution with the selected label. A Score is an ordered semantic value; use its legend before applying numeric thresholds. A Noul is the probability of yes and has no separate confidence field.

Failure cases and production improvements

A regex is a candidate detector, not full email validation. Deduplicate candidates, bound their count, and handle no match explicitly. Do not truncate a long candidate list silently; partition it or require review.

Pin the tested model, log the returned version and rubric revision, and keep a small evaluation set under version control. Re-test after changes to inputs, provider, questions, or policy. Avoid logging sensitive input by default.

Complete code

The downloadable standalone script includes request construction, fixture data, response validation, the decision function, and command-line execution. It uses the direct HTTP API so no SDK dependency is required. For SDK versions of the core ticket request, see Python and JavaScript.

What to test before shipping

• The invoice contact is selected rather than the personal contact.
• The output is an exact supplied candidate, never a generated address.
• No candidate produces an explicit absent/review result.
• A changed candidate list regenerates the criteria map.
• Network failures, invalid JSON, and missing fields must never become an automatic action.
• Compare several thresholds on labeled data and record the resulting review volume.